Development of Room Acoustic Prediction and Evaluation Tools for Designers

Yukiko Seike 1 , Daisaku Miura 2 , Takashi Ishizuka 1 , Kanaya Kitajima 1 , Ayame Kato 1 , Tohru Miyajima 1

1 Institute of Technology, SHIMIZU Corporation 3-4-17 Etchujima, Koto-ku, Tokyo 135-8530, JAPAN 2 Design Division, Building Construction Headquarters, SHIMIZU Corporation 2-16-1 Kyobashi, Chuo-ku, Tokyo 104-8370, JAPAN

ABSTRACT We developed room acoustic prediction and evaluation tools for designers to be used in halls and conference areas where the design of acoustical performance is essential. Designers with no specialized knowledge of acoustics can use the tools to examine the acoustic performance at an early stage of the project. The interface tool consists of 3D CAD software, Rhinoceros, and its plug-in tool, Grasshopper. The designers first build a room model on Rhinoceros and locate the sound source and the receiver points in it. They then set the calculation conditions via Grasshopper, such as room use and type of interior finishes. Next, room sound propagation is calculated using the ray tracing method in external Python programs. Then, acoustic indices such as the average sound absorption coefficient and clarity are predicted. These indices are automatically evaluated according to the room use, and the evaluated results are visualized in the room model on Rhinoceros. This paper outlines these tools and exhibits the agreement of predicted and measured indices in halls. These tools were applied to a lecture room with complicated ceiling ribs and an office with a stairwell, both of which were difficult to examine using the conventional reverberation formula.

1. INTRODUCTION

Acoustic design should be considered in the early stages of a construction project to provide better sound fields, especially in halls, conference areas, and lecture rooms where acoustical performance is essential. Designers must consult an acoustic specialist to study the acoustics properly, but the time and cost burden this involves are high. Hence, designers often start designing acoustics later in the process. In this case, changing the design to ensure or improve the required acoustic performance might be even more difficult or costly. While the general room acoustics prediction tools [1] are very useful for acoustical studies, these require specialized knowledge of acoustics, and designers find these difficult to operate. Therefore, the only way to achieve good acoustic design currently is to consult an acoustic specialist.

To enable designers themselves to confirm the direction of the acoustic design in the early stages of a project and consult with experts easily, we developed room acoustic prediction and evaluation tools for designers using 3D CAD software. This paper outlines these tools and exhibits the agreement of predicted and measured indices in halls. Following this, two case studies and a summary are presented.

inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

2. OVERVIEW OF ROOM ACOUSTIC PREDICTION AND EVALUATION TOOLS

2.1. Aim of the Tools

Figure 1 illustrates the flow of an acoustical study. The conventional study method illustrated in Figure 1 (a) consumes a large amount of time during communication between the designers and acoustic specialists. However, by using the developed tools illustrated in Figure 1 (b), the designers can conduct the acoustic study by themselves and consult with the expert at key points when needed, which makes the acoustic study more efficient.

Designer

Designer

CAD software

Room acoustic simulation and evaluation tools

Create room model

Create room model Setting Calculation

Consult

Evaluation/Judgement Visualization

Acoustic simulation software

Setting Calculation

Visualization

Consult or advise

Evaluation/Judgement

Acoustic specialist

Acoustic specialist

(a) (b)

Figure 1 Flows of an acoustic study: (a) conventional flow, (b) the flow using the developed tools.

Studying acoustics in the early stages of a project has two benefits: 1) a reduction in time and cost for the acoustic study and 2) a high degree of freedom in the design, which improves the acoustic quality. The results of the acoustic study using these tools can determine the general direction for the acoustic design, while an acoustic specialist can be consulted when a more detailed study is required. Because the study by the acoustic specialist is conducted using the same tools, it is continued without interruption.

2.2. Structure of the tools and simulation flow

The tool consists of three parts: “Input room shape,” “Operate,” and “Calculate and evaluate,” as illustrated in Figure 2.

Input room shape

Room shape model

＋ Calculation and Evaluation result Input

Room shape

Rhinoceros

model

Input

Room shape

Operate

data Setting

Grasshopper

Calculation and evaluation result

Visualize Calculation

condition

Run

Calculate and evaluate

Calculation and evaluation result

Acoustic prediction and

evaluation program

Python

Figure 2 Overall tool structure.

inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

In the first part (“Input room shape”), the designers create an inner room shape model which is composed of interior material such as walls, floors, and ceilings using 3D CAD software, Rhinoceros [2]. If room shape data have been created in design work, the model for these tools is easily created based on these data. Components are separated into layers for each type of finishing material. In the same way as actual acoustic measurements, one sound source point and several sound receiving points are set in the model.

Following this, in the second part (“Operate”), users load room shape data with Grasshopper [3], a plug-in tool for Rhinoceros. Grasshopper is a modeling support and visual programming tool that enables algorithm-based design and simulation. Users set several conditions such as material type, use of the room, the number of seats, calculation accuracy level, and so on. The type of material is selected from a list of several dozen types for each layer set in the model, and the sound absorption coefficient and scattering coefficient are set according to the material. The acoustic performances of these materials are automatically read from the database. Parametric studies of material combinations can be performed. There are three calculation accuracy levels: low, medium, and high. After all the settings are complete, all calculation conditions are exported to the acoustic prediction and evaluation programs written in Python, which perform as described in chapter 4.

In the third part (“Calculate and evaluate”), Python programs calculate acoustic prediction, evaluate acoustic physical indices in the background, and return the results to Grasshopper. The results calculated by the program are visualized on the Grasshopper and Rhinoceros models. On the Grasshopper, two types of results for reverberation time calculated by formula and simulation are displayed. The former result is calculated using the Eyring-Knudsen reverberation formula. The latter result displays the average reverberation time of all sound receiving points calculated by the acoustic prediction program. The acoustic physical index values, evaluated results, and reflected sound rays are visualized in Rhinoceros, as illustrated in Figure 3. Figure 3 (a) illustrates a visualized example of acoustic physical index values for each receiving point on the model. The colors of the spheres indicate the evaluated results, which are evaluated according to the rules described in chapter 3. Figure 3 (b) displays the average value or distribution range of all sound receiving points on a graph. Reflected sound rays are visualized as per the sample model illustrated in Figure 3 (c). The color represents the amount of energy in the sound ray when it enters the sound receiving point.

平均吸音率（満席時）

0.24

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

C80 ：音楽明瞭度

1.89[dB]

-5 -4 -3 -2 -1 0 1 2 3 4 5

STI ：音声明瞭度

0.64

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

エコー妨害度 ( 最小・最大 )

10 ～ 50[ ％ ]

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

音圧分布：理論値との差 ( 最小・最大 )

-1.1 ～ -2.1[dB]

-5 -4 -3 -2 -1 0 1 2 3 4 5

(a)

(b) (c)

Figure 3 Examples of visualization of acoustic prediction and evaluation results.

3. EVALUATION OF ACOUSTIC PHYSICAL INDEX

3.1. Evaluation Overview

One of the most important and innovative points of these tools is the automatic evaluation of the acoustic physical indices. Acoustic physical indices to be evaluated and their target ranges are set for each room’s use. By systematizing and automating the evaluation based on the literature and experience of acoustic specialists, designers who do not have professional knowledge can accurately evaluate the results of acoustic prediction calculation.

Table 1 illustrates the list of acoustic physical indices selected for evaluation: average sound absorption coefficient, clarity of music, speech transmission index, echo disturbance, and sound pressure level distribution. These indices are commonly used in room acoustic prediction and calculated from the reflected sound response obtained by acoustic prediction calculation. The values are evaluated as “Good,” “Needs consideration,” or “Bad,” represented by circles, triangles, and crosses, respectively, as illustrated in Figure 3 (b), and scored according to the target range depending on the room use.

Table 1 Configurable room uses and acoustic physical indicators to be evaluated.

Acoustical physical indexes

Speech transmission index

Average sound absorption coefficient

Sound pressure level distribution Concerthall, Multi-purpose hall, Lect ur e hal l (fo r l i v e m usic) 〇 〇 - 〇 〇

Room use

Clarity of music

Echo disturbance

Multi-purpose hall, Live house (electr oa c o us tic s fo r m us i c) 〇 - - 〇 - Multi-purpose hall, Lecture hall, meeting place (electroacoustics for lecture or ceremony) 〇 - - 〇 -

Opera house 〇 〇 - 〇 〇

Church (cathoric) 〇 - - 〇 〇

Church (protestant) 〇 - - 〇 〇

Conference room (midium ～ Large), Class room 〇 - 〇 〇 -

Office 〇 - - - -

Gym, Sports facilities 〇 - - 〇 -

Arena (size for over 10,000 people, ele c tr o a c o us tics fo r m usic) 〇 - - 〇 - Arena (size for over 10,000 people, electroacoustics for Announcement) 〇 - - 〇 -

Atrium (for emergency broadcast communication ） 〇 - - 〇 -

Television studio 〇 - - - -

Radio studio, Announcement booth 〇 - - - -

※　〇： Valuation target ， - ： Out of valuation target

Based on the evaluation of each acoustic physical index determined in the previous section, a comprehensive evaluation, similar to that of acoustic physical indices, is performed. The designer should take the following actions based on the comprehensive evaluation results. If the comprehensive evaluation is “Good,” changes to the room shape and finishes are unnecessary. In the case of “Needs consideration,” the users should change the room shape or finishing materials to improve the acoustic physical indices with a negative evaluation. The “Bad” evaluation means a high probability of major acoustic concerns; so, the room shape or finishing materials must be significantly improved accordingly. If there are difficulties in improving acoustic performance, consulting an acoustic specialist is recommended. As a reference value, the model is evaluated by a total score ranging from 0 to 100.

inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

3.2. Acoustic Physical Indices for Evaluation

The five acoustic physical index values are calculated and evaluated in the following way. The average sound absorption coefficient 𝛼𝛼ത is calculated from the reverberation time 𝑇𝑇 , which in turn, is calculated from the slope of the reverberation decay curve obtained from the reflected sound response as follows: 𝛼𝛼ത = 0.16𝑉𝑉/𝑆𝑆𝑆𝑆 , where 𝑉𝑉 is the room volume, and 𝑆𝑆 is the room surface area.

The clarity of music “C80” is calculated from the ratio of the energy arriving up to 80 ms after the arrival of the direct sound to the energy arriving after that time. The higher the value, the higher the clarity. Based on previous literature [4], we set the recommended range of C80 from -2 to +2 dB for general music clarity in concert halls.

The speech transmission index (STI) is a value that expresses the intelligibility of speech. The degree of sound change from generation to reach is known as the Modulation Transfer Function (MTF), from which STI is calculated [5]. An STI of 0.6 or higher is generally considered desirable in a conversational space such as a conference room [6].

To evaluate echoes, which can obstruct hearing of speech and music, the “% disturbance” proposed by Bolt and Doak is used as an indicator [7]. It is expressed on a scale of 0 to 90%, with a higher level of disturbance represented by a higher value. If the room is used for musical performances, the recommended value is 50%. If it is used for lectures, the recommended value is 30%.

Sound pressure level distribution is used to evaluate the balance of the loudness distribution of sound in the room. The evaluation value is the difference between the value 𝐿𝐿 𝑟𝑟 , calculated by Barron’s modified theoretical equation (4) [8] and the value by the acoustic prediction program. Hence,

𝐿𝐿 𝑟𝑟 = 𝐿𝐿 𝑤𝑤 + 10 log 10 ൤ 1 4𝜋 𝜋 𝑟 𝑟 2 + 𝑇 𝑇 𝑇 13.82𝑉 𝑉 exp ൬− 13.82 𝑇

𝑟 𝑐 𝑐 ൰൨ (4)

where 𝐿𝐿 𝑤𝑤 is the sound source power level, 𝑟𝑟 is the distance between source and receiving points, and 𝑐𝑐 is the sound speed. This formula is based on the theoretical equation assuming a perfectly diffused sound field and considers the distance attenuation of reverberation sound. Figure 4 illustrates the results of the difference between the actual and theoretical values calculated from measured examples. Most points fell within the range of -2dB to +2dB for buildings with highly diffuse shapes and finishes, except when the receiving points were close to sound source points. Therefore, we set the recommended range at -2dB to +2dB.

8

Measured value - Barron's modified theoretical value

6

Sound pressure level

4

difference (dB)

2

0

-2

-4

-6

-8

0 5 10 15 20 25 30 35

Distance from sound source points to sound receiving points (m)

Figure 4 The difference between the measured and Barron’s modified theoretical values.

inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

4. ACOUSTIC SIMULATION AND EVALUATION PROGRAM

The program written in Python consists of two calculation parts. One program performs a geometric acoustic simulation known as the ray tracing method for acoustic prediction calculation. The other one calculates and evaluates various acoustic physical indices using the results calculated by the former program.

Geometric acoustic simulation is commonly used in general-purpose room acoustic simulation software. With the ray tracing method, sound is represented as a particle radiating from a sound source point, and its traces are treated as sound rays. The sound rays have energy, and the decrease in energy stimulates the attenuation of the sound due to sound absorption by reflecting surfaces such as walls, ceilings, and floors and air absorption. In this program, when the sound rays reflect off surfaces, it randomly determines whether each reflection is a specular or scattered reflection based on the scattering coefficient set for each type of finishing material. In the case of scattered reflection, the reflection angle is determined by a random number [9]. Finally, the reflected sound response is drawn as a time series histogram of the time when the sound ray passes through the receiving sphere and the energy of the sound rays. This reflected sound response is treated as an energy expression of the band impulse response obtained in the actual measurement.

From the reflected sound response obtained in the previous paragraph, various acoustic physical index values described in section 3.2 are calculated. Each acoustic physical index value is scored by comparing it to the evaluation criteria. After that, the acoustic performance of the simulated model is comprehensively evaluated based on these scores.

To verify the accuracy of the geometric acoustic simulation program, simulated and measured average reverberation time in several halls were compared. Table 2 illustrates the specifications of the three halls considered here and the results of measurements and simulations. Hall A is a small movie theater with sound-absorbing walls (except in the seating area), while halls B and C were music halls. In all cases, no significant deviations were observed between the measured values and the predicted calculation results. It was confirmed that they demonstrated good correspondence.

Table 2 Overview of the halls considered in sec.4 and measured and simulated average reverberation time.

A B C

Surface area [m 2 ]

991 4,192 6,305 1,268 5,903 12,620 200 1,200 1,500 51 20 353 Measured 0.58 1.74 2.27 Calculated 0.61 1.66 2.41 Reverberation time [s] (500Hz)

Volume [m 3 ] Seats number Measured and Calculated points number

5. CASE STUDY

5.1. Lecture Room

We studied the effectiveness of the tools in a school lecture room with a complex ribbed ceiling. The room was rectangular in shape, with Cross Laminated Timber walls and ceiling, carpet tiles on the floor, and approximately 300 wooden seats. Figure 5 (a) illustrates the ceiling rib shape model. As illustrated in Table 3, four conditions of finishes were compared.

The results indicate that the reverberation times calculated by the Eyring-Knudsen reverberation formula became shorter as the area of sound-absorbing surfaces increased. However, in the acoustic

inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

prediction calculation, the reverberation times shortened only under condition D, where the rear and side wall finishes were absorptive. As illustrated in Figure 5 (b), in the case of a rectangular shape, the reflected sound remains at the opposite reflective surfaces. The reverberation formula based on the statistical theory cannot account for these reflections in the calculation, but the ray tracing method can reproduce them.

inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

(a) (b)

Figure 5 Lecture Room model: (a) ceiling rib shape, (b) remaining sound reflected at opposite reflective surfaces.

Table 3 Reverberation time of lecture room.

Reverberation formula

Aacoustic prediction calculation A Ceilings, Rear walls, Side walls: CLT 5.70 3.74

Finishing Materials

B Rear walls, Side walls: CLT, Ceilings: Perforated plate 1.28 3.89

C Side walls: CLT, Ceilings, Rear walls: Perforated plate 1.03 3.48

D Ceilings: CLT, Rear walls: Perforated plate, Side walls: Curten 0.92 1.16

5.2. Office Building

The distribution of sound pressure levels in an office building with a circular atrium connecting the center of each floor was also examined. Figure 6 (a) illustrates the model used for the study. The circular atrium is designed to overlap on each floor, so there was a concern that the voices on one floor would reflect multiple times in the circular atrium and propagate to the office space on the other floors.

Figure 6 (b) illustrates an example of sound pressure level distribution. The sound source pressure level at a point 1m away is set at 60dB, assuming a talking voice. There were no points where the sound pressure level was especially high in the room. The sound ray diagram, which is not illustrated here, also confirms that no multiple reflections were observed in the circular atrium.

We expect that demands for office acoustics studies like this will increase hereafter. In an office space, the sound diffraction by furniture like partitions and shelves significantly influences sound prediction. Therefore, we plan to develop a calculation method that accounts for diffracted sound in the future. We believe that this planned tool development will be useful for future designs.

inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS. ? O? ? GLASGOW

(a) (b)

Figure 6 Office Building model: (a) overall building elevation, (b) example of sound pressure level distribution. 6. CONCLUSION

We developed a 3D CAD-based room acoustic prediction and evaluation tools for designers who are not acoustic specialists, and the accuracy of the acoustic prediction program was verified. We also applied these tools to a lecture room and an office building as sample studies. This tool enables designers to confirm by themselves the direction of the acoustic design in the early stages of the project and easily consult with experts. Using these tools, designers can reduce the time and cost required for acoustic design and improve acoustic quality.

While these tools were developed mainly for halls and large conference rooms, they can also be used for acoustic analyses of open-plan offices, which have been increasing in recent years. In the future, we will develop a calculation method that takes diffracted sound into account, verify the validity of such an evaluation method, and create a clear auralization of the calculation results. 7. REFERENCES

[1] Odeon A/S: “ODEON,” https://odeon.dk/, Accessed 2022-3-11. [2] AppliCraft Corporation: “Rhinoceros,” https://www.rhino3d.co.jp/, Accessed 2021-10-6. [3] AppliCraft Corporation: “Grasshopper,” https://www.applicraft.com/products/rhinoceros/grasshopper/, Accessed 2021-10-22. [4] V.L. Jordan: “Acoustical Design of Concert Halls and Theatres,” Applied Science Publishers LTD, 1980. [5] Architectural Institute of Japan (ed.): “Evaluation Criteria for Sound Transmission Performance in Urban and Architectural Spaces and Commentary,” November 2011. [6] T. Nakajima and S. Maeda: “‘Ease of Hearing’ and STI of Speech in a Room,” Materials for Architectural Acoustics Research Meeting, 1984. [7] R.H. Bolt and P.E. Doak: “A Tentative Criterion for Short Term Transient Response of Auditoriums,” The Journal of the Acoustical Society of America, 1950. [8] M. Barron, “Energy relations in concert auditoriums I,” The Journal of the Acoustical Society of America, 1988. [9] T. Sakuma: “Small Special Issue - Recent Topics in Room Acoustics - Analysis and Evaluation Method of Diffusivity of Wall Surface,” Journal of the Acoustical Society of Japan, Vol. 61 pp39–44, 2005.

a